Juan Yu has posted comments on this change.

Change subject: IMPALA-3575: Add retry to backend connection request and rpc 
timeout
......................................................................


Patch Set 9:

(1 comment)

http://gerrit.cloudera.org:8080/#/c/3343/9/be/src/runtime/data-stream-sender.cc
File be/src/runtime/data-stream-sender.cc:

PS9, Line 216: ONE_HOUR_IN_MS
> How was one hour chosen? I'm worried this might cause lot of queries to han
yes, that could happen. but we don't want very short timeout here. upstream 
operator sometimes could take very long time and I don't know how long it could 
be. 
This is like the last mean to detect the pair node failure. In most of cases, 
the dest node failure will be detected by statestore heartbeat and the query 
will be cancelled before hitting this timeout.

RPC_TIMEOUT error is checked by IsTimeoutTException(), only if the error 
contains "EAGAIN (timed out)".
It happens only on TSocket::read()


-- 
To view, visit http://gerrit.cloudera.org:8080/3343
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-MessageType: comment
Gerrit-Change-Id: Id6723cfe58df6217f4a9cdd12facd320cbc24964
Gerrit-PatchSet: 9
Gerrit-Project: Impala
Gerrit-Branch: cdh5-trunk
Gerrit-Owner: Juan Yu <[email protected]>
Gerrit-Reviewer: Alan Choi <[email protected]>
Gerrit-Reviewer: Dan Hecht <[email protected]>
Gerrit-Reviewer: Henry Robinson <[email protected]>
Gerrit-Reviewer: Juan Yu <[email protected]>
Gerrit-Reviewer: Sailesh Mukil <[email protected]>
Gerrit-HasComments: Yes

Reply via email to